Tải bản đầy đủ - 0 (trang)
4 Spectral Libraries: State of the Art and Potential Use

4 Spectral Libraries: State of the Art and Potential Use

Tải bản đầy đủ - 0trang

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


Asia, Europe, North America, and South America selected from the Soil Information System (ISIS) of the International Soil Reference and Information

Centre (ISRIC) archives. Samples were scanned in the Vis-NIR spectral

range (350e2500 nm), with a FieldSpec FR spectroradiometer (Analytical

Spectral Devices, Boulder, CO). The soil reference measurements were

acquired by ISRIC in different laboratories according to ISRIC Procedures

for soil analysis (Van Reeuwjik, 2002).

Viscarra Rossel and Webster (2012) described a large library of 21,500

Vis-NIR spectra from around 4000 soil profiles covering the Australian

continent. The soil samples were scanned using an ASD LabSpec Pro spectrometer with a spectral range of 350e2500 nm and spectral resolution of

3 nm at 700 nm and 10 nm at 1400 and 2100 nm. The collection of soil

spectra was realized with a high-intensity contact probe with halogen bulb

illumination. The samples were collected from different soil surveys conducted at different scales (continental, regional, and farm). The soil analyses were

realized in multiple laboratories, following different analysis protocols.

A spectral library covering the United States has been collected under the

Rapid Carbon Assessment project (USDA, 2013). The library is composed

of 144,833 Vis-NIR spectral scans, derived from samples collected from the

upper 1 m of 32,084 soil profiles at 6017 randomly selected locations. The

instrument used to scan the samples was an ASD LabSpec Pro spectroradiometer with a spectral range of 350e2500 nm, 2 nm sampling resolution

and spectral resolution of 3 nm at 700 nm and 10 nm at 1400 and

2100 nm. Soil spectra were acquired using a high-intensity contact probe.

SOC was determined by combustion method.

The European spectral library LUCAS consists of about 20,000 topsoil

(0e20 cm) samples, collected from all over Europe, measured for 13 soil

properties in a single laboratory (Stevens et al., 2013). The Vis-NIR soil

spectra were measured with a FOSS XDS Rapid Content Analyzer

(FOSS NIRSystems Inc., Denmark), operating in the 400e2500 nm wavelength range, with 2 nm spectral resolution and 0.5 nm spectral data interval.

In addition to continental-scale libraries, a number of national and

regional soil spectral libraries have been constructed, such as the ones for

France (Gogé et al., 2012 for NIR and Grinand et al., 2012 for MIR), Czech

Republic (Brodsky et al., 2011), Denmark (Knadel et al., 2012), Florida

(Vasques et al., 2010), and Brazil (Bellinaso et al., 2010).

Soil spectral libraries might be a strong base for the forthcoming hyperspectral remote sensing of soils from space. The laboratory soil spectra may

enable appropriate validation of the reflectance information extracted from


Marco Nocita et al.

radiance data acquired from remote platforms. Moreover soil spectral

libraries can play a major role in tracking temporal spectral changes over

the sampling locations (Deng et al., 2013).

3.5 Soil Spectroscopy for Large-Scale Soil Property


Predicting soil properties for large and diverse areas is especially challenging,

and results in higher prediction error than for local scale spectroscopic

models (Stevens et al., 2013). For instance, Brown et al. (2006) obtained a

root mean square error (RMSE) of 7.9e9.9 g C kgÀ1 for an SOC content

calibration model obtained from samples distributed all over the world. In

Europe, spectral models of SOC content achieved an RMSE of 3.6e

8.9 g C kgÀ1 for mineral soils and 50.6 g C kgÀ1 for organic soils (Nocita

et al., 2014; Stevens et al., 2013). Vasques et al. (2010) developed spectroscopic models of SOC in Florida achieving an RMSE of 6.5e7 g C kgÀ1.

These prediction errors are large compared to the standard error of laboratory (SEL) of established methods of soil carbon analysis, such as dry combustion (SEL ¼ 1e2 g C kgÀ1; Gerighausen et al., 2012). Large-scale

libraries tend to span over a wider range and a higher variability of the

soil property under study, which actually appear to be the dominating factor

influencing prediction errors (Stenberg et al., 2010). The lack of accuracy

shown by spectroscopic models built with large-scale spectral libraries is

also due to the complexity of the relationship between soil properties and

spectra for heterogeneous soil samples. It has been long recognized that

the spectral signature of soils in the Vis-NIR region is not unique (Price,

1994). Many absorption features overlap so that absorptions related to one

soil component can be masked, distorted, or shifted where other soil components vary. For instance, spectral variations related to changes in iron

oxide content may cancel variations in absorptions due to organic matter

(Adar et al., 2014). Not only chemical chromophores interact with each

other. For instance, for the same amount of SOC content, an increase in

sand content induces an increase in SOC absorption depths, which can be

easily confounded with an increase in SOC (Stenberg, 2010; Stevens

et al., 2013). To overcome this, it has been proposed to include additional

variables such as particle size distribution in the modeling of soil spectra

(Brown et al., 2005; Nocita et al., 2014).

The use of large spectral libraries has been proposed also for field and

farm scale soil mapping. In order to overcome the large biases often experienced when using them at the local scale, the spiking technique, consisting

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


in the combination of large-scale libraries with local calibrations, was developed (Guerrero et al., 2010). Basically only a handful of local samples are

included in a large-scale library to shift its weight towards the target site.

In this approach, a few representative samples from the target site (spiking

subset) are added to recalibrate the model, ensuring that the models contain

samples similar to those to predict. This approach implies some analytical efforts, since the spiking subset must be analyzed with the reference method.

Consequently, this subset should be as small as possible, in order to maintain

most of the advantages of spectroscopy.

The poor calibration results at large scales are not only due to the

geographical extent and the resulting complexity as such, but also to the sampling density of the current spectral libraries. Based on the LUCAS spectral

library (w20,000 points across Europe), Nocita et al. (2014) showed, for

instance, that, using a local regression approach (i.e., for each spectrum to

predict, a calibration equation is developed based on the samples with the

most similar spectra or the closest samples in the geographical space), soils

with high SOC content (80 g C kgÀ1) were poorly predicted because of

the lack of suitable nearest neighbors (i.e., samples with similar soil and spectral properties to the ones to predict). Generally, the analyses of such large

databases need dedicated chemometrics tools, such as local regressions that

are able, to some extent, to handle their complexity (Ramirez-Lopez

et al., 2013). Development costs of large databases are high, so that such spectral libraries are often developed from archived soil samples and legacy soil

databases with analytical measures (Viscarra Rossel and Webster, 2012).

Although these libraries contain an enormous wealth of information on soils,

they cannot be easily merged into a uniform database because they have been

collected with different protocols, instruments, and analytical methods

which can severely affect the prediction performance of spectroscopic

models (Soriano-Disla et al., 2014). For instance, Brown et al. (2005)

computed an RMSE of 6 g C kgÀ1 for 1175 samples analyzed for SOC

both by dry combustion and the Walkley-Black method. Such error would

be included in the total error budget of SOC spectroscopic models based on

samples analyzed with the two different methods. Obviously this represents a

waste of resources since most spectral libraries cannot be exploited together

to create robust models over large areas and with diverse soil types.

3.6 Parameters Causing Spectral Variation in the Laboratory

Not only the soil components but also the laboratory protocols have an influence on the spectra. Depending on the instruments, samples are prepared


Marco Nocita et al.

following a procedure specific to each laboratory, which may result in difference in spectral shapes for the same sample between different laboratories.

Often, soil samples are dried, sieved, and grinded. Differences in water content of air-dried samples, due to fluctuations in relative humidity of the

ambient air in the laboratory, affect the spectral shape and peaks, especially

around 1.415 and 1.915 mm (Whiting et al., 2004; Nocita et al., 2012).

Spectral reflectance is also affected by the grinding of soil. This can generate

important differences of accuracy in the prediction models due to the variation of particle sizes (Soriano-Disla et al., 2014). Unfortunately, there is no

protocol specifying at which caliber samples should be grinded. The same is

true for sieving. This makes the construction of comparable spectral libraries

difficult, and precludes the sharing of spectra among laboratories with

different measurement conditions and protocols. Pimstein et al. (2011)

showed that the use of a common protocol and an internal standard reduced

significantly the differences between spectral measurements of the same samples by different operators in three laboratories.

The comparison of analytical results among laboratories is traditionally

addressed by performing ring tests with a standard sample being sent and

analyzed by all participating laboratories. For spectral analysis, it is recommended to send both a reference material, such as bleached inert sand,

and a standard soil sample to the laboratories. The spectra of these materials

can then be used to determine transfer function for the spectra determined in

one laboratory to be used in a spectral library of another laboratory.

3.7 Metadata and Soil Spectroscopy

Metadata are structured information that make an information resource

easier to access, use, and understand. Just as for any other kind of information derived in the field or the laboratory, the utility of soil spectroscopy

data is only as good as how well it can be explained and understood.

Capturing the metadata at the observation stage is a lot more cost-effective

than trying to work out all of the necessary information later on, and can aid

in the integration of the data within global frameworks, and its later extraction and use.

Many different definitions and frameworks for metadata have been

developed, reflecting the wide variety of data and information types that

can be generated. Existing examples of metadata standards of relevance to

soil spectroscopy include ISO 19115 (ISO, 2003), the Dublin Core Metadata Initiative (DMCI, 2013), and the Directory Interchange Format

(NASA, 2013).

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


The seven questions that the notion of metadata addresses are all relevant

to soil spectroscopy, and are listed below:

• What was measured? This can include the spectroscopic information,

additional environmental parameters, and soil analysis results obtained

later in the lab.

• Who carried out the measurements? Some field observations and laboratory analyses are operator-dependent, and it can be useful to know who

carried out the work in order to calibrate for this (and also to be able to

ask them questions about the data if the need arises).

• Where were the measurements made? Geographical location of field observations can allow later site characterization information to be derived.

• When were the measurements made? This information is particularly

important when long-term monitoring or changes over time are of


• How were the measurements made? Type of equipment, field sampling

protocols any other seemingly innocuous information about how the

data were derived can be useful later.

• Why was the work carried out? Was it in relation to some other project,

and were the observations intended to satisfy some specific requirement?

• Whose is the data? Intellectual property is a topic that tends to be ignored

until it becomes an unavoidable issuedhaving information about

ownership, rights of use, and referencing of the material available to

the user early on can facilitate license negotiations and prevent problems

occurring once work has been carried out.

Suggestions for implementation of metadata standards within soil spectroscopy are likely to cover a wide range of topics, from the adoption of

preferred standards to the inclusion of specific types of information (and

the formatting of the metadata framework). There is a great deal of flexibility

available in how such a system could be adopted, and we are not attempting

here to prescribe how this should be achieved. However, any successful

metadata framework should aim at the very least to be compliant with the

INSPIRE directive (http://inspire.ec.europa.eu/), and it is proposed that

the information and guidance available in relation to this directive would

be a suitable starting point for discussions.


The growing demand for high-resolution soil data to cover large areas

on the one hand, and the lack of availability of such data on the other hand,


Marco Nocita et al.

is one of the biggest challenges in contemporary soil science (Grunwald

et al., 2011), encouraging thereby the development of cost-effective

methods of soil analysis, such as Vis-IR spectroscopy. While research on

soil spectroscopy has rapidly grown and showed a great potential (Guerrero

et al., 2010), it is now time for soil spectroscopy to enter an operational phase

where, just as for other established soil analytical techniques, measurements

are standardized, soil analyses are reliable across diverse environments and

data are delivered in an automated mechanism. To achieve this goal, one

prerequisite is the development of databases that can provide robust spectral

models over large geographical extents. It is unlikely that the high sampling

density, required to appropriately describe soil variation at these scales, could

be reached by a single research group. However, the combination of existent

and future local, regional, and continental spectral libraries is an achievable

target, provided that they are built using a common protocol for the collection of laboratory soil spectra, or that they contain spectral reference measurements so that spectral transfer functions could be calculated. Similarly

to spectral measurements, the use of reference analytical methods that spectroscopic predictions rely on should be standardized. Obviously, a joint

effort of the soil spectroscopy community is required to allow better interoperability between soil spectral libraries and facilitate data exchange. Such

initiative is not only a way to unlock current limitations of soil spectroscopy,

but also a lever that would directly promote the use of spectroscopy in academic and commercial soil laboratories, favoring the development of

monitoring networks able to assess quickly and efficiently the state of the

soil resources at minimal costs. In the next sections, we present some ideas

that, once implemented, could contribute to the expansion of soil spectroscopy as an established soil analytical method.

4.1 Establishment of a Common Protocol for Laboratory


Since calibration and management represent a large part of costs and efforts

in the use of IR techniques, standardization and centralization of reference

methods are strongly needed. This represents the first step for the sharing of

small- and large-scale spectral libraries, which could help in achieving calibration models that are valid for larger areas. So far, all the national and continental spectral libraries have been built using slightly different protocols for

each library including sampling technique, sample preparation, instrument

specifications, and spectral acquisition, which hinder interoperability of

spectral libraries. The use of different reference methods and the problems

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


related with quality control of reference measurements within and across

laboratories provide major challenges for the development of reliable calibrations. Iterative development of centralized spectral libraries is efficiently

achieved by screening spectral libraries and then conducting reference measurements on outlying samples. An international standard for the collection

of laboratory spectra and the inclusion of spectra of reference materials will

dramatically drop the costs linked with the collection of new samples. Moreover, the development of a common standard, together with a common

network for scientists and technicians from all over the world, will give

rise to new applications, such as the transfer of calibration models from laboratory to spectra of the soil surface collected by remote sensing.

4.2 Scanning of Existent Soil Archives

Sampling campaigns are costly, and soil archives stored by universities,

research centers, agriculture associations, and government agencies could

provide an opportunity to enlarge spectral libraries. Even if many samples

were acquired decades ago, they still contain spectral information that could

be used to improve the representativity of spectroscopic calibration models.

These samples could be scanned to a cost basically corresponding to the

working hours of a technician. In many cases, however, considerations on

the property of reference data and on the confidentiality of some metadata

(especially location) have to be addressed. A common framework could be

proposed for negotiating with data owners (e.g., access to spectra, and

imprecise geographical coordinates). The impact of this mobilization, realized following a common protocol, might be part of the answer to the demand for robust calibration model across regions and soil types. Since 2006,

ISO/TC 190 (soil quality)/SC 3/WG 10 (screening methods) has been

developing standards to screen soil for chemical compounds including heavy

metals, petroleum, the total carbon, and nitrogen as well as harmful anions

such as chromium (VI) and cyanides, under the guidance established in 2011

as ISO 12404 on screening methods to be applied to soil monitoring.

4.3 Storing Spectra and Associated Soil Archives

The spectrum of a soil sample contains abundant information. Once

scanned, a spectrum can be stored easily. It is probable that, in the future,

soil scientists might be measuring some properties which at the present

time are not measured, because neither the knowledge nor appropriate

analytical techniques are available. The characteristics of the soils stored as

spectra will remain unaltered, while soil properties will change after a long


Marco Nocita et al.

storage. This is why soil spectral libraries should be accompanied with soil

banks that conserve the soil samples to be scanned or analyzed in the future.

4.4 Spectroscopy to Acquire Standardized Soil Information

and Enhance Monitoring

The implementation of soil spectroscopy could find valuable application in

the soil information system and would represent a great progress in the field

of soil analysis. For instance, soil classification can be realized by several

methods (Soil Survey Staff, 1999; IUSS Working Group WRB, 2006),

but the high costs for its implementation has hindered the collection of

this information. Soil spectroscopy can be applied for the classification of

soil types (Viscarra Rossel and Webster, 2012). Demattê and Terra (2014)

demonstrated the potential of spectroscopy for the evaluation of the changes

in soil type along topo-sequences, as a basic tool for soil mapping. Moreover,

Demattê et al. (2004) showed that soil spectroscopy can be used to map soil

types, as the basis for land-use planning.

At the same time, soil spectroscopy could be integrated and thus regulate

the use of fertilizers based on routine soil testing. For example, one of the

main problems of a reliable fertility status is the lack of knowledge of the

CEC and clay content of our soils. Indeed, those properties are the key to

compare samples to the regional reference system and deliver an appropriate

agronomic diagnosis. The determination of CEC and clay content using

standard procedures is not feasible for each soil sample as it is too costly

and time-consuming. An alternative is to predict them with Vis-NIR spectroscopy which represents thereby a real opportunity to improve the fertility

advice (Genot et al., 2011). In Wallonia, these analyses are routinely carried

out since 2008 in the laboratories of the http://www.requasud.be/. Another

example of soil spectroscopy application for routine analysis is the AfSIS.

They have adopted Vis-IR spectroscopy as its main screening tool in characterizing 20,000 soil samples taken from a stratified random sampling frame

across Sub-Saharan Africa. There is a growing network of Vis-IR laboratories in Africa, with more than 10 laboratories established, including two

private sector soil testing companies.

Recently soil spectroscopy was reported as an accurate method to

monitor temporal changes in SOC of Danish soils (Deng et al., 2013). Basically, this study used topsoil (0e25 cm) samples from 1986 to 2009. Spectra

collected from both time series revealed that Vis-NIR spectroscopy could

soundly detect the SOC temporal decrease observed with wet chemistry,

but at much lower cost.

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


In conclusion, given the adherence to a common protocol, spectroscopic analyses could both increase the reliability and the comparability

of the results and, at the same time, contribute to the construction of

the spectral libraries. The latter requires quality assessment/quality control

steps to be built in the database for the input of new spectra. Diffusing the

use of Vis-NIR and MIR spectrometers at all levels, especially farmers’

consortia and agriculture organizations, responsible for the soil analyses

of thousands of samples collected over large areas, could bring important

advantages. The long-term cost cuts from which these associations would

benefit will no doubt justify the investment in instruments. Many regional

laboratories already operate Vis-NIR spectrometers to infer soil properties.

This implies that the establishment of local and regional partnerships

among existent institutions will already generate enough competence

for the soil monitoring based on soil spectra. The development of this

kind of project requires the support of international organizations, such

as the Food and Agriculture Organization (FAO) of United Nations and

the European Commission’s JRC, and their acknowledgment of soil spectroscopy as a valuable tool to integrate the established techniques of soil

chemical analyses for the control of the state of soils (Clark and Roush,



 ., 2012. Predicting soil

Abdi, D., Tremblay, G.F., Ziadi, N., Bélanger, G., Parent, L.-E

phosphorus-related properties using near-infrared reflectance spectroscopy. Soil Sci.

Soc. Am. J. 76, 2318e2326.

Adar, S.Y., Shkolnisky, Y., Ben Dor, E., 2014. Change detection of soils under small-scale

laboratory conditions using imaging spectroscopy sensors. Geoderma 216, 19e29.

Aenugu, H.P.R., Sathis Kumar, D., Srisudharson, P.N., Ghosh, S.S., Banji, D., 2011. Near

infrared spectroscopy- an overview. Int. J. ChemTech Res. 3, 825e836.

Alpert, N.L., Keiser, W.E., Szymanski, H.A., 1970. Theory and Practice of Infrared

Spectroscopy. Plenum Press, New York.

Bellinaso, H., Dematte, J.A.M., Araujo, S.R., 2010. Spectral library and its use in soil

classification. Braz. J. Soil Sci. 34, 861e870.

Bellon-Maurel, V., McBratney, A., 2011. Near-infrared (NIR) and mid-infrared (MIR)

spectroscopic techniques for assessing the amount of carbon stock in soils e critical

review and research perspectives. Soil Biol. Biochem. 43, 1398e1410.

Ben-Dor, E., Banin, A., 1995. Near-infrared analysis as a rapid method to simultaneously

evaluate several soil properties. Soil Sci. Soc. Am. J. 59, 364e372.

Ben Dor, E., Inbar, Y., Chen, Y., 1997. The reflectance spectra of organic matter in the

visible near-infrared and short wave infrared region (400e2500 nm) during a controlled

decomposition process. Remote Sens. Environ. 61, 1e15.

Ben Dor, E., Irons, J.R., Epema, J.F., 1999. Soil reflectance. In: Manual of Remote Sensing:

Remote Sensing for the Earth Sciences, vol. 3. John Wiley & Sons, New York,

pp. 111e188.


Marco Nocita et al.

Ben Dor, E., Patkin, K., Banin, A., Karnieli, A., 2002. Mapping of several soil properties using DAIS-7915 hyperspectral scanner data. A case study over clayey soils in Israel. Int. J.

Remote Sens. 23, 1043e1062.

Ben Dor, E., Chabrillat, S., Dematte, J.A.M., Taylor, G.R., Hill, J., 2009. Using imaging

spectroscopy to study soil properties. Remote Sens. Environ. 113, 38e55.

Bornemann, L., Welp, G., Amelung, W., 2010. Particulate organic matter at field scale: rapid

acquisition using mideinfrared spectroscopy. Soil Sci. Soc. Am. J. 74, 1147e1156.

Bowers, S., Hanks, R.J., 1965. Reflectance of radiant energy from soils. Soil Sci. 100,


Brodsky, L., Klement, A., Penizek, V., Kodesova, R., Boruvka, L., 2011. Building soil spectral library of the Czech soils for quantitative digital soil mapping. Soil Water Res. 6,


Brown, D.J., Bricklemyer, R.S., Miller, P.R., 2005. Validation requirements for diffuse

reflectance soil characterization models with a case study of VNIR soil C prediction

in Montana. Geoderma 129, 251e267.

Brown, D.J., Shepherd, K.D., Walsh, M.G., Dewayne Mays, M., Reinsch, T.G., 2006.

Global soil characterization with VNIR diffuse reflectance spectroscopy. Geoderma

132, 273e290.

Calderon, F.J., Reeves III, J.B., Collins, H.P., Eldor, P.A., 2011. Chemical differences in soil

organic matter fractions determined by diffuse-reflectance mid-infrared spectroscopy.

Soil Sci. Soc. Am. J. 75, 568e579.


nasveras Sanchez, J., Barr

on, V., Del Campillo, M., Viscarra Rossel, R.A., 2012. Reflectance spectroscopy: a tool for predicting soil properties related to the incidence of Fe

chlorosis. Span. J. Agric. Res. 10, 1133e1142.

Clark, R.N., Roush, T.L., 1984. Reflectance spectroscopy: Quantitative analysis techniques

for remote sensing applications. J. Geophys. Res. 89, 6329e6340.

Dalal, R.C., Henry, R.J., 1986. Simultaneous determination of moisture, organic carbon,

and total nitrogen by near infrared reflectance spectrophotometry. Soil Sci. Soc. Am.

J. 50, 120e123.

Davis, T., 1998. The history of near infrared spectroscopic analysis: past, present and

future - from sleeping technique to the morning star of spectroscopy. Anal. Mag.

26, 17e19.

Demattê, J.A.M., Campos, R.C., Alves, M.C., Fiorio, P.R., Nanni, M.R., 2004. VisibleNIR reflectance: a new approach on soil evaluation. Geoderma 121, 95e112.

Demattê, J.A.M., Terra, F., 2014. Spectral pedology: a new perspective on evaluation of soils

along pedogenetic alterations. Geoderma 217e218, 190e200.

Demyan, M.S., Rasche, F., Schulz, E., Breulmann, M., Muller, T., Cadisch, G., 2012. Use of

specific peaks obtained by diffuse reflectance Fourier transform mid-infrared spectroscopy to study the composition of organic matter in a Haplic Chernozem. Eur. J. Soil

Sci. 63, 189e199.

Deng, F., Minasny, B., Knadel, M., McBratney, A., Heckrath, G., Greve, M.H., 2013. Using

vis-NIR spectroscopy for monitoring temporal changes in soil organic carbon. Soil Sci.

178, 389e399.

Dublin Core Metadata Initiative, 2013. http://dublincore.org/.

Eisele, A., Lau, I., Hewson, R., Carter, D., Wheaton, B., Ong, C., Cudahy, T.J.,

Chabrillat, S., Kaufmann, H., 2012. Applicability of the thermal infrared spectral region

for the prediction of soil properties across semi-arid agricultural landscapes. Remote Sens.

4, 3265e3286.

Genot, V., Colinet, G., Bock, L., Vanvyve, D., Reusen, Y., Dardenne, P., 2011. Nearinfrared reflectance spectroscopy for estimating soil characteristics valuable in the diagnosis of soil fertility. J. Near Infrared Spectrosc. 19, 117e138.

Soil Spectroscopy: An Alternative to Wet Chemistry for Soil Monitoring


Gerighausen, H., Menz, G., Kaufmann, H., 2012. Spatially explicit estimation of clay and

organic carbon content in agricultural soils using multi-annual imaging spectroscopy

data. Appl. Environ. Soil Sci. 23.

Gobrecht, A., Roger, J.M., Bellon-Maurel, V., 2014. Major issues of diffuse reflectance NIR

spectroscopy in the specific context of soil carbon content estimation: a review. Adv.

Agron. 123, 145e175.

Gogé, F., Joffre, R., Jolivet, C., Ross, I., Ranjard, L., 2012. Optimization criteria in sample

selection step of local regression for quantitative analysis of large soil NIRS database.

Chemom. Intell. Lab. 110, 168e176.

Grinand, C., Barthes, B.G., Brunet, D., Kouakoua, E., Arrouays, D., Jolivet, C., Caria, G.,

Bernoux, M., 2012. Prediction of soil organic and inorganic carbon contents at a national

scale (France) using mid-infrared reflectance spectroscopy (MIRS). Eur. J. Soil Sci. 63,


Grunwald, S., Thompson, J.A., Boettinger, J.L., 2011. Digital soil mapping and modeling at

continental scales e finding solutions for global issues. Soil Sci. Soc. Am. J. 75, 1201e1213.

Guerrero, C., Zornoza, R., Gomez, I., Mataix-Beneyto, J., 2010. Spiking of NIR regional

models using samples from target sites: effect of model size on prediction accuracy. Geoderma 158, 66e77.

He, J., Song, H., García Pereira, A., Hernandez G

omez, A., 2005. A new approach to predict

N, P, K and OM content in a loamy mixed soil by using near infrared reflectance

spectroscopy. Adv. Intell. Comput. 3644, 859e867.

Herzberg, G., 1945. Molecular Spectra and Molecular Structure: Infrared and Raman Spectra

of Polyatomic Molecules. D Van Nostrand, New York.

International Organization for Standardisation (ISO), 2003. ISO 19115. http://www.iso.


Islam, K., Singh, B., McBratney, A., 2003. Simultaneous estimation of several soil properties by ultra-violet, visible, and near-infrared reflectance spectroscopy. Soil Res. 41,


IUSS Working Group WRB, 2006. World Reference Base for Soil Resources, second ed.

FAO, Rome. World Soil Resources Reports No. 103.

Janik, L.J., Skjemstad, J.O., 1995. Characterization and analysis of soils using mid-infrared

partial least-squares. II. Correlations with some laboratory data. Austr. J. Soil Res. 33,


Knadel, M., Deng, F., Thomsen, A., Greve, H.M., 2012. Development of a Danish National

Vis-NIR Soil Spectral Library for Soil Organic Carbon Determination. Digital Soil Assessments and beyond. CRC Press, Boca Raton, FL, pp. 403e408.

Louis, B.P., Saby, N.P.A., Orton, T.G., Lacarce, E., Boulonne, L., Jolivet, C., Ratié, C.,

Arrouays, D., 2014. Statistical sampling design impact on predictive quality of harmonization functions between soil monitoring networks. Geoderma 213, 133e143.

Malley, D.F., Martin, P.D., Ben-Dor, E., 2004. Application in analysis of soils. In:

Roberts, C.A., Workman Jr., J., Reeves III, J.B. (Eds.), Near-Infrared Spectroscopy in

Agriculture. American Society of Agronomy, Crop Science Society of America, Soil

Science Society of America, pp. 729e784.

McCarty, G.W., Reeves III, J.B., Reeves, V.B., Follett, R.F., Kimble, J.M., 2002. MidInfrared and near-infrared diffuse reflectance spectroscopy for soil carbon

measurement. Soil Sci. Soc. Am. J. 66, 640e646.

NASA, 2013. Directory Interchange Format Writer’s Guide. http://gcmd.gsfc.nasa.gov/


Nocita, M., Stevens, A., Noon, C., Van Wesemael, B., 2012. Prediction of soil organic carbon for different levels of soil moisture using Vis-NIR spectroscopy. Geoderma 199,


Tài liệu bạn tìm kiếm đã sẵn sàng tải về

4 Spectral Libraries: State of the Art and Potential Use

Tải bản đầy đủ ngay(0 tr)